{
 "cells": [
  {
   "cell_type": "code",
   "execution_count": null,
   "id": "9de5907f-18f5-4cb1-903e-26028ff1fa03",
   "metadata": {},
   "outputs": [],
   "source": [
    "from CountryCountsByBatchcodeTable2BarChartDescriptionTableConverter import CountryCountsByBatchcodeTable2BarChartDescriptionTableConverter\n",
    "from CountriesByBatchcodeProvider import getCountryCountsByBatchcodeTable, filterByBatchcodes\n",
    "from Column2DataframeAdder import addColumn2Dataframe\n",
    "from BatchCodeTableFactory import BatchCodeTableFactory\n",
    "from HistogramFactoryAndPersister import createAndSaveGlobalHistograms\n",
    "from SymptomByBatchcodeTableFactory import SymptomByBatchcodeTableFactory\n",
    "from BatchCodeTablePersister import createGlobalBatchCodeTable\n",
    "from BatchCodeTableHtmlUpdater import updateBatchCodeTableHtmlFile, saveLastUpdatedBatchCodeTable\n",
    "from InternationalVaersCovid19Provider import getInternationalVaersCovid19, get_international_VAERSVAX_VAERSSYMPTOMS_Covid19\n",
    "from DateProvider import DateProvider\n",
    "from datetime import datetime\n",
    "from VAERSFileDownloader import updateVAERSFiles\n",
    "from IOUtils import IOUtils\n",
    "import os\n",
    "import pandas as pd\n",
    "\n",
    "pd.set_option('display.max_rows', 100)\n",
    "pd.set_option('display.max_columns', None)\n"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "id": "d1e4fa9e",
   "metadata": {},
   "outputs": [],
   "source": [
    "print(datetime.now().strftime(\"%d.%m.%Y, %H:%M:%S Uhr\"))\n"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "id": "ffad1c04",
   "metadata": {},
   "outputs": [],
   "source": [
    "dateProvider = DateProvider()\n",
    "print('          lastUpdated:', dateProvider.getLastUpdated())\n",
    "print('lastUpdatedDataSource:', dateProvider.getLastUpdatedDataSource())\n",
    "needsUpdate = dateProvider.needsUpdate()\n",
    "print('needsUpdate:', needsUpdate)\n"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "id": "194b7357",
   "metadata": {},
   "outputs": [],
   "source": [
    "years_from_start_of_COVID_vaccination_to_present = list(range(2020, datetime.now().year + 1))\n"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "id": "a793dff0",
   "metadata": {},
   "outputs": [],
   "source": [
    "updateVAERSFiles(\n",
    "        years = years_from_start_of_COVID_vaccination_to_present,\n",
    "        workingDirectory = os.getcwd())"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "id": "58333a19",
   "metadata": {},
   "outputs": [],
   "source": [
    "international_VAERSVAX_Covid19, international_VAERSSYMPTOMS = get_international_VAERSVAX_VAERSSYMPTOMS_Covid19(years = years_from_start_of_COVID_vaccination_to_present)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "id": "f10b558f",
   "metadata": {},
   "outputs": [],
   "source": [
    "international_VAERSVAX_Covid19\n"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "id": "3e0908fe",
   "metadata": {},
   "outputs": [],
   "source": [
    "international_VAERSSYMPTOMS\n"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "id": "781ac80e",
   "metadata": {},
   "outputs": [],
   "source": [
    "internationalVaersCovid19 = getInternationalVaersCovid19(dataDir = 'VAERS', years = years_from_start_of_COVID_vaccination_to_present)\n",
    "internationalVaersCovid19"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "id": "0915aa5a",
   "metadata": {},
   "outputs": [],
   "source": [
    "batchCodeTable = createGlobalBatchCodeTable(\n",
    "    deleteEntriesWithADRsLessThanOrEqual = 2,\n",
    "    minADRsForLethality = 100,\n",
    "    batchCodeTableFactory = BatchCodeTableFactory(internationalVaersCovid19))\n",
    "batchCodeTable"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "id": "d8e81ffc",
   "metadata": {},
   "outputs": [],
   "source": [
    "IOUtils.saveDataFrameAsJson(batchCodeTable, '../docs/data/batchCodeTables/Global.json')\n",
    "saveLastUpdatedBatchCodeTable(\n",
    "    DateProvider().getLastUpdatedDataSource(),\n",
    "    batchCodeTableHtmlFile=\"../docs/batchCodes.html\")"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "id": "189a639e",
   "metadata": {},
   "outputs": [],
   "source": [
    "updateBatchCodeTableHtmlFile(batchCodeTable, batchCodeTableHtmlFile=\"../docs/HowBadIsMyBatch.html\")"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "id": "19fa050b",
   "metadata": {},
   "outputs": [],
   "source": [
    "symptomByBatchcodeTable = SymptomByBatchcodeTableFactory.createSymptomByBatchcodeTable(\n",
    "    VAERSVAX = international_VAERSVAX_Covid19[international_VAERSVAX_Covid19['VAX_LOT'].isin(batchCodeTable['Batch'])],\n",
    "    VAERSSYMPTOMS = international_VAERSSYMPTOMS)\n",
    "symptomByBatchcodeTable"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "id": "ec2a692e",
   "metadata": {},
   "outputs": [],
   "source": [
    "createAndSaveGlobalHistograms(symptomByBatchcodeTable, batchCodeTable)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "id": "70fcc420",
   "metadata": {},
   "outputs": [],
   "source": [
    "countryCountsByBatchcode = filterByBatchcodes(getCountryCountsByBatchcodeTable(), batchCodeTable['Batch'].values)\n",
    "countryCountsByBatchcode"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "id": "7a67b1f6",
   "metadata": {},
   "outputs": [],
   "source": [
    "barChartDescriptionTable = CountryCountsByBatchcodeTable2BarChartDescriptionTableConverter.convert2BarChartDescriptionTable(countryCountsByBatchcode)\n",
    "barChartDescriptionTable"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "id": "7ebfd53c",
   "metadata": {},
   "outputs": [],
   "source": [
    "def dateRange2Str(dateRange):\n",
    "    return {\n",
    "        'start': _date2Str(dateRange[0]),\n",
    "        'end': _date2Str(dateRange[1])\n",
    "    }\n",
    "\n",
    "def _date2Str(date):\n",
    "    return date.strftime(\"%d.%m.%Y\")"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "id": "0ec101b6",
   "metadata": {},
   "outputs": [],
   "source": [
    "from GoogleAnalyticsReader import GoogleAnalyticsReader\n",
    "\n",
    "dateRange = GoogleAnalyticsReader(dataDir = 'data/GoogleAnalytics').getDateRange()\n",
    "barChartDescriptions = {\n",
    "    'dateRange': dateRange2Str(dateRange),\n",
    "    'barChartDescriptions': barChartDescriptionTable['BAR_CHART_DESCRIPTION'].to_dict()\n",
    "}\n"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "id": "163830ba",
   "metadata": {},
   "outputs": [],
   "source": [
    "IOUtils.saveDictAsJson(barChartDescriptions, '../docs/data/barChartDescriptionTable.json')\n"
   ]
  },
  {
   "attachments": {},
   "cell_type": "markdown",
   "id": "2d93b511",
   "metadata": {},
   "source": [
    "# Vaccine Distribution by Zipcode"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "id": "cfcbad44",
   "metadata": {},
   "outputs": [],
   "source": [
    "vaccineDistributionByZipcode  = pd.read_excel(\n",
    "        io = 'data/Amended-22-01962-Pfizer-2022-0426-pulled-2022-0823.xlsx',\n",
    "        usecols = ['PROVIDER_NAME', 'ZIPCODE_SHP', 'LOT_NUMBER', 'DOSES_SHIPPED'],\n",
    "        dtype = {'DOSES_SHIPPED': 'int'})\n",
    "vaccineDistributionByZipcode"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "id": "59c745d2",
   "metadata": {},
   "outputs": [],
   "source": [
    "from VaccineDistributionByZipcodeSimplifier import VaccineDistributionByZipcodeSimplifier\n",
    "\n",
    "vaccineDistributionByZipcode = VaccineDistributionByZipcodeSimplifier.sumDoses(vaccineDistributionByZipcode)\n",
    "vaccineDistributionByZipcode"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "id": "8cd250f7",
   "metadata": {},
   "outputs": [],
   "source": [
    "OVERALL_DOSES_SHIPPED_by_LOT_NUMBER = vaccineDistributionByZipcode.groupby('LOT_NUMBER').agg(OVERALL_DOSES_SHIPPED = pd.NamedAgg(column = 'DOSES_SHIPPED', aggfunc = sum))\n",
    "OVERALL_DOSES_SHIPPED_by_LOT_NUMBER"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "id": "1a5667be",
   "metadata": {},
   "outputs": [],
   "source": [
    "vaccineDistributionByZipcode = addColumn2Dataframe(dataframe = vaccineDistributionByZipcode, column = OVERALL_DOSES_SHIPPED_by_LOT_NUMBER)\n",
    "vaccineDistributionByZipcode"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "id": "f77505c6",
   "metadata": {},
   "outputs": [],
   "source": [
    "from ADR_by_Batchcode_Table_Factory import create_ADR_by_Batchcode_Table_4USA\n",
    "\n",
    "ADR_by_Batchcode_Table_4USA = create_ADR_by_Batchcode_Table_4USA(internationalVaersCovid19)\n",
    "ADR_by_Batchcode_Table_4USA"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "id": "99120c77",
   "metadata": {},
   "outputs": [],
   "source": [
    "vaccineDistributionByZipcode = addColumn2Dataframe(dataframe = vaccineDistributionByZipcode, column = ADR_by_Batchcode_Table_4USA)\n",
    "vaccineDistributionByZipcode"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "id": "3276cce7",
   "metadata": {},
   "outputs": [],
   "source": [
    "vaccineDistributionByZipcode['Statistical Number of Adverse Reaction Reports'] = (vaccineDistributionByZipcode['DOSES_SHIPPED'] / vaccineDistributionByZipcode['OVERALL_DOSES_SHIPPED'] * vaccineDistributionByZipcode['Adverse Reaction Reports']).round(2)\n",
    "vaccineDistributionByZipcode['Statistical Number of Adverse Reaction Reports (per 100,000)'] = (vaccineDistributionByZipcode['DOSES_SHIPPED'] / vaccineDistributionByZipcode['OVERALL_DOSES_SHIPPED'] * 100000).round().astype(int)\n",
    "vaccineDistributionByZipcode"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "id": "10cf731f",
   "metadata": {},
   "outputs": [],
   "source": [
    "vaccineDistributionByZipcode = vaccineDistributionByZipcode[['PROVIDER_NAME', 'ZIPCODE_SHP', 'LOT_NUMBER', 'DOSES_SHIPPED', 'Statistical Number of Adverse Reaction Reports', 'Statistical Number of Adverse Reaction Reports (per 100,000)']]\n",
    "vaccineDistributionByZipcode"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "id": "0c2020e9",
   "metadata": {},
   "outputs": [],
   "source": [
    "vaccineDistributionByZipcode = vaccineDistributionByZipcode.rename(\n",
    "    columns = {\n",
    "        'PROVIDER_NAME': 'Provider',\n",
    "        'ZIPCODE_SHP': 'ZIP Code',\n",
    "        'LOT_NUMBER': 'Lot Number',\n",
    "        'DOSES_SHIPPED': 'Doses Shipped'\n",
    "    })\n",
    "vaccineDistributionByZipcode"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "id": "f68c72d0",
   "metadata": {},
   "outputs": [],
   "source": [
    "# vaccineDistributionByZipcode.to_excel('tmp/Amended-22-01962-Pfizer-2022-0426-pulled-2022-0823_sumDoses.xlsx')\n"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "id": "9b8f0b6e",
   "metadata": {},
   "outputs": [],
   "source": [
    "from IOUtils import IOUtils\n",
    "\n",
    "IOUtils.saveDataFrameAsJson(vaccineDistributionByZipcode, '../docs/data/vaccineDistributionByZipcode/VaccineDistributionByZipcode.json')"
   ]
  }
 ],
 "metadata": {
  "kernelspec": {
   "display_name": "howbadismybatch-venv-kernel",
   "language": "python",
   "name": "howbadismybatch-venv-kernel"
  },
  "language_info": {
   "codemirror_mode": {
    "name": "ipython",
    "version": 3
   },
   "file_extension": ".py",
   "mimetype": "text/x-python",
   "name": "python",
   "nbconvert_exporter": "python",
   "pygments_lexer": "ipython3",
   "version": "3.9.15"
  },
  "vscode": {
   "interpreter": {
    "hash": "1bce2b9b19ce5f16d695ff75ac05095b3e564c169ff454b58b87cb796c0695b8"
   }
  }
 },
 "nbformat": 4,
 "nbformat_minor": 5
}